[X3D-Public] HTML5 Web3D Working Group Dec 29 2009

Tue Jan 5 14:27:59 PST 2010

X3D-Public mailing list members - here are the finalized minutes for  
the December 29 2009 meeting.

What: 		Web3D HTML5 meeting December 29 2009
present: 		John Stewart, Johannes Behr, Joe Williams, Don Brutzman,  
Anita Havelle, Leonard Daly.
absent: 		

Web page: 	http://www.web3d.org/x3d/wiki/index.php/X3D_and_HTML5
-----------------------------------------------------------------------------------------------------------

----------------------------------------------------------------------------------------------------------------------------------------------
3) W3C Bug 8238: Sam Ruby and list of X3D Elements and Attributes.

http://www.w3.org/Bugs/Public/show_bug.cgi?id=8238

This list contains abstract types, MF and SF enumerations, and other  
keywords that need not be captured by HTML5.

ACTION - Don - will autogenerate element and attribute lists from the  
schema and send it along to the HTML5 wg, in response to bug 8238.

November 24: no change.

December  1: no change.

December 8: no change.

December 29:

Don - updated and took ownership of bug. Schema and DTD fixed in bug  
listing.

Don - will keep bug open; Don will maintain bug.

Joe - do we support interchange, interactive, etc profiles or  
everything?

Don - we are going for full profile.

DONE.

----------------------------------------------------------------------------------------------------------------------------------------------
4) Case independent encoding of X3D.

Question: Should we have an encoding that closely matches HTML5?

If we do, HTML5 content model -  our goal would be that if we have an  
all lower case encoding that a parser that reads in a relaxed parsing  
mode, we would promote the capitalization for the DOM model interaction.

Noted that in X3D, there are no ambiguities created by ignoring case;  
eg, there are no conflicts for "Interpolator" and "interpolator".

Also noted that HTML5, case does not matter, but the DOM is case  
sensitive.

ACTION: All - do we relax case on xml parsing in general. (background  
- html5 is case insensitive; can we cut and paste)

November 24:

Don - case insensitive is their terms for "lax".

Don - what the spec says for SVG and MathML will tell us what we  
should do. Don will go through spec on upcoming trip overseas; see  
below. Returns 19th December;

DONE ACTION: Johannes - x3dom.org, make parser relaxed in terms of  
case sensitivity. Changed x3dom.org implementation and added example  
with mixed up case encoding.

December 1
Johannes - work on html pages with no namespaces (thus, not xhtml) -  
it is possible with the opengl implementation. This is from requests  
from people who wish to autogenerate content.

Don - who takes care of case conversion? eg, do the html parsers  
actually give the content to web3d browsers in the textual case  
received, or do they convert the case to upper (or) lower case?

ACTION:  Johannes; will verify with known and unknown content. Google  
Chrome, Webkit and Firefox nightly build.

Don - our tools should strive to be flexible with input, but also  
ensure that they provide strictly correct output.

December 8:
Johannes - the x3dom implementation was originally designed and built  
on XHTML (which is case sensitive and namespace aware) but we are  
porting it work on non-namespace-aware, case-insensitive HTML.  Our  
initial investigation has led us to the point where we are not really  
sure - it looks like some HTML browsers behave differently than others  
in terms of what is sent to x3dom- but looks like all browsers store  
DOM in lower case.  This task is still ongoing - this is just a  
preliminary report.

December 29:
Johannes - it will be case insensitive if HTML; case sensitive if in  
XHTML.

DONE

----------------------------------------------------------------------------------------------------------------------------------------------
5) General access to HTML5 group minutes.

The minutes are sent to members of this group, and to the public-x3d  
mailing list. They are also archived to the Web3D internal archives,  
but should be easily located by the general public.

ACTION: JohnS - add wiki section, and add hypermail links to minutes.

November 24: not done yet. Had difficulty accessing Web3D.org wiki  
from inside work firewall.

Don - hint: to do this, left square bracket, url, space, prose "oct  
minutes right square bracket.

December 1: JohnS - access problems to web3d wiki experienced last  
week, will resolve issues and put more time into it this week.

Len; Do we want to broadcast to the world that we have in-process, or  
should some discussions remain private?

Don - This question is worth revewing. Either we are open or we are  
not. Not sure we could claim openness is we were not publishing what  
we are doing.

Don - Also have not had a problem with non-members contributing yet.  
Hopefully we will get to that problem.

Don - As part of review cycle the minutes go through us first for  
verification.

DONE.

----------------------------------------------------------------------------------------------------------------------------------------------
6) Discussion on Void element (comment 6 in bug 8238)

November 24:

http://www.w3.org/Bugs/Public/show_bug.cgi?id=8238

hint: the only contained content in any element is eg: the javascript  
url code in a script node (Shaders, too)

Don: Void element - talking about a singleton element where they have  
omitted the closing slash at the end of the tag.

Joe - void element became what xml calls empty element.

Johannes - has definition... http://dev.w3.org/html5/markup/syntax.html#syntax-elements

Don - every element can have metadata child, so that is a problem.  
Also, there are a handful of things that are not an element; like ROUTE.

Don - we can give a list of nodes that have metadata and nothing  
else....

Joe - HTML in principle does not recognize the self closing tag. And  
yet, if you put that in all the browsers;  the browsers are NOT  
rejecting it as invalid content.

Don - Example: if we have 3 transforms in a row; how do you tell if  
they are nested children or just siblings if not closing tags?

Don - do we have any examples of void tags in html?

Joe - Yes, void tags - metadata, parameters,....

Don - do html5 have a list of void tags for html; if they do not have  
a list, then we have to give one back to them.

Johannes - void elements must not have an end tag.

joe - if you put a <br/>, all tested browsers will not mind.

Johannes - but, this is not spec - it is a relaxing of spec input  
parsing rules.

Joe - but the browsers are ok with this. so void is the same as empty  
in both html and xhtml.

joe - so in a way, the emphasis should be on html5 be ok on the  
closing tag.

ACTION: Joe to send a couple of examples of this.

December 1 - no action.

December 8:
JohnS:  will put on hold until html5 browsers get more settled. I will  
keep the topic header in minutes, but will remove previous text until  
topic has been reopened.

----------------------------------------------------------------------------------------------------------------------------------------------
7) x3dom X3D integration issues.

Johannes - x3dom.org testing; , expose javascript object part that  
maps to fields to try to map to SAI so that one can do faster updates  
rather than ASCII-ized DOM, as per what the SVG people have done.  
Would like to implement this and test to determine efficiency of  
asking the DOM element for the javascript object so you can manipulate  
this directly.

One nice thing is that one can use all the X3D definitions of the SAI  
for the fields; can spec a sub part of SAI spec for this.

Noted: SVG currently requires this methodology for performance and  
convenience.

Don - so, need DOM Extensions for X3D spec.

Don - the more we do this, the more we reduce the barriers of entry.  
This is why Sam Ruby (W3C) wants to pose us as a case for central  
extensibility vs decentralized extensibility.

Johannes - might be able to encode as JSON;

Joe, JSON is just a transport kind of thing.....

Johannes; have to do InstantReality release, will work on this x3dom  
extension over christmas.

December 1 : no progress.

December 7 :
Johannes - next major x3dom item in the development queue is a task  
that will to expose the interfaces in x3dom.

December 29:
Johannes - will send an example that will make this clear when he is  
back at the office.

----------------------------------------------------------------------------------------------------------------------------------------------

8) Updates to HTML5 spec.

forked from Item 4) above, the following paragraph:

         "Don - what the spec says for SVG and MathML will tell us  
what we should do. Don will go
          through spec on upcoming trip overseas; see below. Returns  
19th December".

December 1: No change.

December 8: No change.

December 29: Don had laptop problems - getting closer on task; maybe  
next week task will be finalized.

----------------------------------------------------------------------------------------------------------------------------------------------

9) External SAI Scripting tests.

December 1:

Don Brutzman by email:  "we need to get some external SAI examples  
that show
javascript in web page talking to an embedded X3D scene."

December 8:
JohnS - FreeWRL uses SAI externally from Java, C, and Tcl (ie, not  
embedded in a Script node). The
Javascript classes have been "blocked in" but code has not been tested  
nor made active in releases yet.

December 29:

Joe - mailed an SAI example from external script.

All to review.

Don - wants examples of routing of events via SAI.

Len - 2 years ago, Vivaty supports sai; now support is not supported.  
Bitmanagement and Octaga have some form of SAI.

Don - whether we have implementations or not, if we have an example or  
two, we can use it to ask companies "hey where is our scripting  
interface?"

Len - SAI spec does not tell how to start the connection so that each  
browser can implement SAI differently.

Johannes - if we put HTML4 example out there, we need to put the HTML5  
equivalent - so do we want HTML4 examples?

Don - we need a set of examples that work in either environment. - we  
are asking our companies to do the right thing when X3D is encountered.

Joe, Don, Len: Examples are being distributed via email. Examples will  
be formalized by next meeting.

----------------------------------------------------------------------------------------------------------------------------------------------

10) 3d-test.com interview.

December 8:
JohnS - We have been asked by Benoit Vandangeon to participate in an  
on-line interview for
http://www.3d-test.com. We will discuss this by email, and will return  
a group-derived answer by 20 December.

December 29:
Anita - news item on Web3D.org based on interview.

JohnS - Javascript comment and precision issues found by x3dom - if we  
have an floating point issue; we will work with the javascript people  
to get issues resolved.

Johannes; Don; Johns: Javascript by default supports doubles.

Don - here is the text from the Ecmascript spec outlining that all  
numbers are double:

Standard ECMA-262
3r d Edition - December 1999
ECMAScript Language Specification

===================
4.3.19 Number Value
A number value is a member of the type Number and is a direct  
representation of a number.

4.3.20 Number Type
The type Number is a set of values representing numbers. In ECMAScript,
the set of values represents the double-precision 64-bit format IEEE 754
values including the special “Not-a-Number” (NaN) values, positive  
infinity,
and negative infinity.
===================

JohnS: but, most GPUs only support 32 bit internally.

Joe: some are 64 bit now:

JohnS: very true, but they are not widely distributed, like on pdas,  
etc.

JohnS: so two potential javascript issues:
	1) there are potential 64 <--> 32 bit conversions being performed at  
all gpu interactions;
	2) if Javascript uses OpenCL or CUDA for calculations, it will have  
precision difficulties.

-----------------------------------------------------------------------------------------------------------------------------------------------

-----------------------------------------------------------
John A. Stewart
Team Leader: Networked Virtual and Augmented Reality
alex.stewart at crc.ca

Network Systems and Technologies -
         Systemes et technologies des reseaux
Communications Research Centre Canada  |
          Centre de recherches sur les communications Canada

3701 Carling Ave.  |  3701, avenue Carling
PO Box 11490, Station H  |  CP 11490, succursale H
          Ottawa ON K2H 8S2   |  Ottawa (Ontario) K2H 8S2

http://www.crc.ca